Update README.md
Browse files
README.md
CHANGED
|
@@ -143,8 +143,10 @@ model_name = "deepseek-ai/DeepSeek-V3.1-Terminus"
|
|
| 143 |
model, tokenizer,_=llm_load_model(model_name,trust_remote_code=False,device="cpu")
|
| 144 |
layer_config = {}
|
| 145 |
for n, m in model.named_modules():
|
|
|
|
|
|
|
| 146 |
if isinstance(m, torch.nn.Linear):
|
| 147 |
-
if
|
| 148 |
layer_config[n] = {"bits": 8}
|
| 149 |
continue
|
| 150 |
if "expert" in n and "shared_experts" not in n:
|
|
|
|
| 143 |
model, tokenizer,_=llm_load_model(model_name,trust_remote_code=False,device="cpu")
|
| 144 |
layer_config = {}
|
| 145 |
for n, m in model.named_modules():
|
| 146 |
+
if isinstance(m,torch.nn.Embedding):
|
| 147 |
+
layer_config[n] = {"bits": 8}
|
| 148 |
if isinstance(m, torch.nn.Linear):
|
| 149 |
+
if n=="lm_head":
|
| 150 |
layer_config[n] = {"bits": 8}
|
| 151 |
continue
|
| 152 |
if "expert" in n and "shared_experts" not in n:
|